jrrribeiro commited on
Commit
75d15f3
·
1 Parent(s): 4a9f781

feat(ui): implement audio loading and cache cleanup functionality in Gradio app

Browse files
README.md CHANGED
@@ -50,3 +50,4 @@ Estrutura base criada por projeto/dataset:
50
  - Sprint 1: CLI create-project, sync-audio, build-index, verify-project
51
  - Sprint 2 (inicial): fila paginada de deteccoes com filtros basicos em demo local Gradio
52
  - Sprint 2 (backend): AudioFetchService com download sob demanda por audio_id + cache efemero com cleanup pos-validacao
 
 
50
  - Sprint 1: CLI create-project, sync-audio, build-index, verify-project
51
  - Sprint 2 (inicial): fila paginada de deteccoes com filtros basicos em demo local Gradio
52
  - Sprint 2 (backend): AudioFetchService com download sob demanda por audio_id + cache efemero com cleanup pos-validacao
53
+ - Sprint 2 (UI): carregar audio da deteccao selecionada e limpar cache apos validacao
src/ui/app_factory.py CHANGED
@@ -1,10 +1,25 @@
1
  import gradio as gr
 
2
 
 
3
  from src.domain.models import Detection
4
  from src.repositories.in_memory_detection_repository import InMemoryDetectionRepository
 
5
  from src.services.detection_queue_service import DetectionQueueService
6
 
7
 
 
 
 
 
 
 
 
 
 
 
 
 
8
  def _seed_service() -> DetectionQueueService:
9
  repo = InMemoryDetectionRepository()
10
  demo_items = [
@@ -70,12 +85,65 @@ def _page_to_table(service: DetectionQueueService, page: int, scientific_name: s
70
  return rows, status, page_obj.page
71
 
72
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
  def create_app() -> gr.Blocks:
74
  service = _seed_service()
 
75
 
76
  with gr.Blocks(title="BirdNET Validator HF") as demo:
77
  gr.Markdown("# BirdNET Validator HF")
78
- gr.Markdown("Sprint 2 kickoff: fila paginada de deteccoes (demo local).")
 
 
79
 
80
  with gr.Row():
81
  species_filter = gr.Textbox(label="Filtro especie", placeholder="Ex: Cyanocorax cyanopogon")
@@ -92,6 +160,14 @@ def create_app() -> gr.Blocks:
92
  label="Deteccoes",
93
  interactive=False,
94
  )
 
 
 
 
 
 
 
 
95
  status = gr.Textbox(label="Status", interactive=False)
96
 
97
  def refresh(page: int, species: str, confidence: float):
@@ -103,6 +179,13 @@ def create_app() -> gr.Blocks:
103
  def go_prev(page: int, species: str, confidence: float):
104
  return refresh(max(1, page - 1), species, confidence)
105
 
 
 
 
 
 
 
 
106
  demo.load(
107
  fn=refresh,
108
  inputs=[page_state, species_filter, min_confidence],
@@ -123,5 +206,22 @@ def create_app() -> gr.Blocks:
123
  inputs=[page_state, species_filter, min_confidence],
124
  outputs=[table, status, page_state],
125
  )
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
126
 
127
  return demo
 
1
  import gradio as gr
2
+ from typing import Protocol
3
 
4
+ from src.cache.ephemeral_cache_manager import EphemeralCacheManager
5
  from src.domain.models import Detection
6
  from src.repositories.in_memory_detection_repository import InMemoryDetectionRepository
7
+ from src.services.audio_fetch_service import AudioFetchService
8
  from src.services.detection_queue_service import DetectionQueueService
9
 
10
 
11
+ class _AudioFetchResultProtocol(Protocol):
12
+ cache_key: str
13
+ local_path: str
14
+ source: str
15
+
16
+
17
+ class _AudioServiceProtocol(Protocol):
18
+ def fetch(self, dataset_repo: str, audio_id: str) -> _AudioFetchResultProtocol: ...
19
+
20
+ def cleanup_after_validation(self, cache_key: str) -> None: ...
21
+
22
+
23
  def _seed_service() -> DetectionQueueService:
24
  repo = InMemoryDetectionRepository()
25
  demo_items = [
 
85
  return rows, status, page_obj.page
86
 
87
 
88
+ def _extract_audio_id(rows: object, selected_index: int) -> str:
89
+ normalized_rows: list[list[object]]
90
+
91
+ if hasattr(rows, "values"):
92
+ normalized_rows = [list(item) for item in rows.values.tolist()]
93
+ else:
94
+ normalized_rows = [list(item) for item in rows] if rows else []
95
+
96
+ if not normalized_rows:
97
+ raise ValueError("Nenhuma deteccao carregada na tabela")
98
+ if selected_index < 0 or selected_index >= len(normalized_rows):
99
+ raise ValueError("Selecione uma deteccao valida na tabela")
100
+
101
+ value = normalized_rows[selected_index][1]
102
+ audio_id = str(value).strip()
103
+ if not audio_id:
104
+ raise ValueError("audio_id invalido na deteccao selecionada")
105
+ return audio_id
106
+
107
+
108
+ def _fetch_selected_audio(
109
+ audio_service: _AudioServiceProtocol,
110
+ dataset_repo: str,
111
+ rows: object,
112
+ selected_index: int,
113
+ previous_cache_key: str,
114
+ ) -> tuple[str | None, str, str]:
115
+ repo = dataset_repo.strip()
116
+ if not repo:
117
+ return None, "", "Informe dataset repo no formato owner/repo"
118
+
119
+ try:
120
+ audio_id = _extract_audio_id(rows=rows, selected_index=selected_index)
121
+ result = audio_service.fetch(dataset_repo=repo, audio_id=audio_id)
122
+ status = f"Audio carregado ({result.source}) para audio_id={audio_id}"
123
+ return result.local_path, result.cache_key, status
124
+ except Exception as exc:
125
+ if previous_cache_key:
126
+ return None, previous_cache_key, f"Falha ao carregar audio: {exc}"
127
+ return None, "", f"Falha ao carregar audio: {exc}"
128
+
129
+
130
+ def _cleanup_selected_audio(audio_service: _AudioServiceProtocol, cache_key: str) -> tuple[str, str | None]:
131
+ if not cache_key:
132
+ return "Nenhum audio em cache para limpar", None
133
+
134
+ audio_service.cleanup_after_validation(cache_key=cache_key)
135
+ return "Cache de audio limpo apos validacao", None
136
+
137
+
138
  def create_app() -> gr.Blocks:
139
  service = _seed_service()
140
+ audio_service = AudioFetchService(EphemeralCacheManager(ttl_seconds=300, max_files=128))
141
 
142
  with gr.Blocks(title="BirdNET Validator HF") as demo:
143
  gr.Markdown("# BirdNET Validator HF")
144
+ gr.Markdown("Sprint 2: fila paginada + audio sob demanda com cache efemero.")
145
+
146
+ dataset_repo = gr.Textbox(label="Dataset repo", value="SEU_USUARIO/birdnet-projeto-dataset")
147
 
148
  with gr.Row():
149
  species_filter = gr.Textbox(label="Filtro especie", placeholder="Ex: Cyanocorax cyanopogon")
 
160
  label="Deteccoes",
161
  interactive=False,
162
  )
163
+ selected_index = gr.Number(label="Linha selecionada", value=0, precision=0)
164
+
165
+ with gr.Row():
166
+ load_audio_btn = gr.Button("Carregar audio selecionado")
167
+ clear_audio_btn = gr.Button("Limpar cache apos validacao")
168
+
169
+ audio_player = gr.Audio(label="Audio sob demanda", type="filepath")
170
+ cache_key_state = gr.State(value="")
171
  status = gr.Textbox(label="Status", interactive=False)
172
 
173
  def refresh(page: int, species: str, confidence: float):
 
179
  def go_prev(page: int, species: str, confidence: float):
180
  return refresh(max(1, page - 1), species, confidence)
181
 
182
+ def on_select(evt):
183
+ if isinstance(evt.index, tuple):
184
+ return int(evt.index[0])
185
+ if isinstance(evt.index, int):
186
+ return int(evt.index)
187
+ return 0
188
+
189
  demo.load(
190
  fn=refresh,
191
  inputs=[page_state, species_filter, min_confidence],
 
206
  inputs=[page_state, species_filter, min_confidence],
207
  outputs=[table, status, page_state],
208
  )
209
+ table.select(fn=on_select, inputs=None, outputs=[selected_index])
210
+ load_audio_btn.click(
211
+ fn=lambda repo, rows, idx, cache_key: _fetch_selected_audio(
212
+ audio_service=audio_service,
213
+ dataset_repo=repo,
214
+ rows=rows,
215
+ selected_index=int(idx),
216
+ previous_cache_key=cache_key,
217
+ ),
218
+ inputs=[dataset_repo, table, selected_index, cache_key_state],
219
+ outputs=[audio_player, cache_key_state, status],
220
+ )
221
+ clear_audio_btn.click(
222
+ fn=lambda cache_key: _cleanup_selected_audio(audio_service=audio_service, cache_key=cache_key),
223
+ inputs=[cache_key_state],
224
+ outputs=[status, audio_player],
225
+ )
226
 
227
  return demo
tests/unit/test_app_factory_audio_helpers.py ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from dataclasses import dataclass
2
+ from pathlib import Path
3
+
4
+ import pandas as pd
5
+
6
+ from src.ui.app_factory import _cleanup_selected_audio, _extract_audio_id, _fetch_selected_audio
7
+
8
+
9
+ @dataclass
10
+ class FakeFetchResult:
11
+ cache_key: str
12
+ local_path: str
13
+ source: str
14
+
15
+
16
+ class FakeAudioService:
17
+ def __init__(self) -> None:
18
+ self.cleaned: list[str] = []
19
+
20
+ def fetch(self, dataset_repo: str, audio_id: str) -> FakeFetchResult:
21
+ _ = dataset_repo
22
+ return FakeFetchResult(cache_key=f"key:{audio_id}", local_path=f"/tmp/{audio_id}.wav", source="remote")
23
+
24
+ def cleanup_after_validation(self, cache_key: str) -> None:
25
+ self.cleaned.append(cache_key)
26
+
27
+
28
+ def test_extract_audio_id_from_list_rows() -> None:
29
+ rows = [["k1", "audio_01", "sp", 0.9, 0.0, 1.0]]
30
+ assert _extract_audio_id(rows, 0) == "audio_01"
31
+
32
+
33
+ def test_extract_audio_id_from_dataframe_rows() -> None:
34
+ frame = pd.DataFrame([["k1", "audio_02", "sp", 0.9, 0.0, 1.0]])
35
+ assert _extract_audio_id(frame, 0) == "audio_02"
36
+
37
+
38
+ def test_fetch_selected_audio_success() -> None:
39
+ service = FakeAudioService()
40
+ rows = [["k1", "audio_03", "sp", 0.9, 0.0, 1.0]]
41
+
42
+ path, cache_key, status = _fetch_selected_audio(
43
+ audio_service=service,
44
+ dataset_repo="org/dataset",
45
+ rows=rows,
46
+ selected_index=0,
47
+ previous_cache_key="",
48
+ )
49
+
50
+ assert path == "/tmp/audio_03.wav"
51
+ assert cache_key == "key:audio_03"
52
+ assert "Audio carregado" in status
53
+
54
+
55
+ def test_fetch_selected_audio_validates_repo() -> None:
56
+ service = FakeAudioService()
57
+ rows = [["k1", "audio_03", "sp", 0.9, 0.0, 1.0]]
58
+
59
+ path, cache_key, status = _fetch_selected_audio(
60
+ audio_service=service,
61
+ dataset_repo=" ",
62
+ rows=rows,
63
+ selected_index=0,
64
+ previous_cache_key="old-key",
65
+ )
66
+
67
+ assert path is None
68
+ assert cache_key == ""
69
+ assert "Informe dataset repo" in status
70
+
71
+
72
+ def test_cleanup_selected_audio() -> None:
73
+ service = FakeAudioService()
74
+
75
+ status, player_value = _cleanup_selected_audio(service, "key:audio_03")
76
+
77
+ assert "Cache de audio limpo" in status
78
+ assert player_value is None
79
+ assert service.cleaned == ["key:audio_03"]